Assessing Mixing Sensitivities for Scale-up

Matthew Jörgensen

Atlanta, 10/20/2014

Nalas: Solutions that Scale!

Engineering Services

Broad range of services & capabilities supporting chemical process development…

Outline

Interplay of scale up and mixing with Examples at Nalas • Example 1: Identification of the critical mixing

parameters • Example 2: Modeling the mixing of the process at

both the large and small scale – Requiring multiple experiments to match all scales of

mixing

• Example 3: Modeling the mixing throughout the process – As the process changes, so can the mixing

Mixing at Scale or Scale of Mixing? • Definition of what we mean by “Scale” regarding mixing?

– The process scale, i.e. lab, pilot, production

– Mixing scale, i.e. micro mixing, mesomixing, macro mixing

• We Need to take both definitions of “scale” into account for the process

What is the Scale of the Mixing?

• Micro-mixing is mixing on the smallest scales of motion (the Kolmogorov scale)

• Mesomixing refers to the turbulent dispersion of a feed stream shortly after it enters a mixing vessel. – It is caused by the action of the bulk fluid interacting with the feed stream. – Mesomixing occurs at a higher scale compared to micromixing, but at a lower

scale than macromixing.

• Macromixing, i.e. the blend time in a batch system • Which one is important?

– What time scale is your process?

Micro: Fast reactions Meso: Semi-batch process

Macro: Bulk blending

Example 1: Nitration Quench

• Product is synthesized by mixed acid nitration • Material is isolated by Quenching into water

– Crystallization is instantaneously upon addition to water

• Varying levels of residual acid have been detected in lots of product by IC – Larger scale batches have higher acid levels – Residual acid has been shown to be due to inclusion in

the solids, i.e. not surface acid that can be washed off.

Example 1: Nitration Quench

Statement of the problem:

• Level of mixing present in the quench seems to correlate with acid levels

• How do we design the quench so as to minimize the levels of acid inclusion

In order to Scale up, we need to scale down!

Critical Mixing Parameter(s)?

• Need to identify the mixing parameters that are controlling – The first step was to run reactions with various

degrees of mixing

• Identify a process responds – In this case levels of acid inclusion.

• Find the mixing parameters that correlated with the responds – The rapid crystallization points to micro-mixing time

being critical

Scale down Mixing for Quench

Reactor RPM Characteristic time of micromixing

100-L Large Scale 200 4.84 seconds

100-mL EasyMaxTM with probes 1000 0.69 seconds

100-mL EasyMaxTMwith probes 500 1.58 seconds

100-mL EasyMaxTM w/o probes 320 4.92 seconds

100-mL EasyMaxTM with probes 185 4.86 seconds

100 ml reactor 100 L reactor

With VisiMix it was possible to adjust the degree of baffling and rpms to match the micro mixing time in the lab to the large scale reactor

Relationship of Quench Mixing and Acid Inclusion from Lab Experiments

Experiment RPMs MicroMixing time [sec] Sulfuric acid Nitric Acid

“Poor” mixing 320 4.92 0.556% 0.083%

“Intermediate” mixing 500 2.43 0.084% 0.058%

“good’ mixing 1000 0.691 0.022% 0.055%

Multiple mixing outputs from VisiMix were compared with the Acid levels

Micromixing time correlated best with the data As all reactions had the

same feed time Data would also correlate with mesomixing time

Large scale experiments with slower dosing rates did not reduce acid levels this indicats that

micromixing time is the correct factor that dictates acid inclusion level

Example 1: Summary

• First identify the critical mixing parameter(s) – Select a process responds (quality attribute, yield, etc.) – Correlate a mixing parameter with this responds

• Use VisiMix to design the lab scale reaction to match the critical mixing parameter(s) of the large scale equipment – The “poor” mixing in the lab equipment was the best that could

be achieved at the 100L scale

• Use the data from the lab runs to scale up the process – With the correlation of micromixing time and acid levels, it was

possible to design an optimal quenching approach • The optimal quench is a small scale continuous quench • Allows for the required micromixng time to prevent acid inclusion

Example 2: Scale-up of an Energetic, Scale-down of the Mixing

• The reactions to form intermediate 2 are very rapid – intermediate 1 is never seen in the reaction.

• Intermediate 2 is above the solubility limit for majority of the reaction. – As the reaction is done within the meta-stable zone solids can form.

• Reactions to form product becoming mass transfer limited further slowed formation of product. – Product is not stable in reaction mixture

NH2NH2

NH2N

+ N

NH2NO2

N+

NO2

N

O2N NO2

Intermediate 1

Intermediate 2

Intermediate 3

ProductStarting material

fast

fast

slow

Possible solid

Mass transfer limited

Not stable in reaction mixture

Mixing and Concentrations • Depending on the Meso-mixing the local concentration can be higher

• very high local concentrations could rapidly form solids • Mixing could narrow the MSZW due to high shear, resulting in solids

Solution Solid

MSZW narrowing due to mixing1

Possible process concentration Intermediate 2

Possible process concentration due to poor Meso-Mixing

Temperature

Co

nce

ntr

atio

n

1. Chianese, A., Contaldi, A. and Mazzarotta, B., J Crystal Growth,1986, 78: 279–290.

*Cartoon data to highlight point

MSZW

Questions to answer: Is Intermediate 2 going to precipitate out? Are the solids going to reduce the yield?

Evaluation of the mixing

• Evaluate the mixing in the 100 gallon reactor – Need to match the micro mixing time, meso

mixing time, and the max energy dissipation and shear

• It will not be possible to match all three in one experiment – We will run three reactions, each matching one of

the mixing parameters.

– Evaluate each reaction for solid formation and impact on reaction rates and yield

Scale Down of the Mixing: VisiMix Modeling of 100 Gallon and Lab Reactor

100 gal 0.5L Tr,Cal, Raman, pH 0.5L Tr,Cal, Raman, pH 0.5L Tr,Cal, Raman, pH

175 rpms 935 rpms 535 rpms 300 rpms

Volume at end of dose 81 gal 0.35 L 0.35 L 0.35 L

Max energy [W/kg] 73.9 74 13.9 2.45

bulk energy [W/kg] 0.193 1.030 0.194 0.0341

average energy [W/kg] 0.624 2.07 0.389 0.0686

micromixing time [s] 2.28 0.983 2.27 4.83

Shear rate in: [1/s]

bulk 440 1020 442 185

near impeller 8630 8640 3740 1570

tip speed [m/s] 3.37 1.84 1.05 0.589

Mixing power [W] 191 0.726 0.136 0.024

Reynolds # for flow 3.28E+05 13400 7640 4280

Avg tangential velocity [m/s] 0.802 0.233 0.133 0.0744

Avg circulation velocity [m/s] 0.0842 0.296 0.169 0.0949

Feed rate[m^3/sec] 2.20E-05 2.50E-08 2.50E-08 1.32E-07

[mL/min] 1320.00 1.50 1.50 7.92

k (Turbulent diffusivity) 6.81E-01 2.03E-01 6.62E-02 2.08E-02

Dt (energy dissipation rate) 2.41E-01 4.01E-03 2.26E-03 1.27E-03

Td(Meso mixing time) [sec] 1.09E-03 2.11E-05 6.56E-05 1.09E-03

Dose time [min] 90.15 90.00 90.00 17.05

Baldyga and Bourne mesomixing time for dispersion of feed1 D=QB/UDt

QB=Volumetric feed rate U= Fluid velocity in surrounding fluid at feed point Dt= local turbulent diffusivity =0.1k2/ and k~0.06U2

tip

1. Edward Paul(ed),Handbook of Industrial Mixing Science and Practice, 2004, John Wiley & Sons. ISBN: 0-471-26919-0. p772.

Reaction formed solids

Matching Micro and Meso Mixing,

During the dose, no solids seen Next morning, no solids seen No Mass Transfer Limitation for Reactions

No mass transfer limitation seen in Intermediate 2 profile, In situ Yield of 71% at 18 hour reaction time

Intermediate 2

Intermediate 3 Product

Reagent

In Situ Raman Spectroscopy for Reaction Matching Micro-Mixing

Matching Max Energy Dissipation

End of Dose solids seen, but not “gummy”

Next morning, no solids seen Mass transfer limited Reaction

Mass transfer limitation seen in Intermediate 2 profile, In situ Yield of 68% at 18 hour reaction time

Intermediate 2 solid Intermediate 3 Product

Reagent Intermediate 2 solution

In Situ Raman Spectroscopy for Reaction Matching Max Energy Dissipation

Mixing and Concentrations

• Meso-mixing local concentration will not lead to solid formation

Solution Solid

MSZW narrowing due to mixing1

Possible process concentration Intermediate 2

Temperature

Co

nce

ntr

atio

n

1. Chianese, A., Contaldi, A. and Mazzarotta, B., J Crystal Growth,1986, 78: 279–290.

*Cartoon data to highlight point

MSZW

Max sheer and energy dissipation near impeller can narrow MSZW

Example 2: Summary

• Not possible to match all the mixing parameters at each scale in one reaction – Requires multiple experiments to isolate each

mixing parameter of interest

• Possible Mixing impact on process, the chance exists for solids to form in the 100 gallon reactor – Due to max energy dissipation and shear near

impeller • The solids will not be problematic for the process and

would result in only a slight yield loss, 68% vs. 71%.

Case study 3: Non-Newtonian Fluid and Mixing Throughout the Process • Customers is scaling up a deprotection of a polymer

solution. – Solution is a non-Newtonian fluid that changes throughout

the process

• Wanted to know if their process would perform at the 1000 gallon scale – Process had already been scaled to the 275 gallon scale

• Needed to evaluate additional process parameters (total amount of off gas) – Wanted to scale down the mixing in the lab equipment to

be sure the data would be representative of the large scale equipment.

Non-Newtonian fluids • A Non-Newtonian fluid is one where the viscosity is not constant with shear rate.

• In this case the fluid is shear thinning (viscosity decreases as shear increases)

• Viscosity as a function of sheer rates was supplied by costumer as well as densities at three points in the process

• This data was fit to a power-law model for viscosity in VisiMix

– The Herschel-Bulkley form of the equation.

Non-Newtonian Fluid Viscosities

• The values for the parameters in the power-law were regressed to allow for modeling viscosity over the range of shear rates seen in the tank

K n τ˳Time zero 6.63E-02 9.62E-01 8.18E-02

Phase transition 1.00E-05 5.00E-02 6.03E+00

Final 2.51E-01 3.94E-01 0.00E+00

Highly viscous with no vortex at this point

Viscosity dramatically decreases, over 100 ml of volume increase due to vortex

Visimix ® Model of reactors

• The geometry for the RC-1, the 275 gallon reactor, and the 1000 gallon reactor were input into Visimix ®

200 gallon fill (275 gallon reactor)

737 gallon fill (1000 gallon reactor)

0.1 gallon fill (0.4 L) (0.26 gallon reactor)

What Mixing Parameters are Important?

• In practice it is not possible to match all the mixing parameters at various process scales.

– rpms, tip speed, etc.

• Important to identify parameters that describe and impact the process.

– Comparison of apparent viscosity at different zones in the tank (cP)

– Average energy dissipation (W/kg)

Visimix ® Results Time Zero

275 gallon R-203 @110 RPMs

1000 gallon R-1002 @90 RPMs

RC-1 Pitch Blade @750 RPMs

Viscosity bulk volume (cP) 53.8 53.8 Viscosity near baffles (cP) 50.1 51.7 50.4 Viscosity near impeller (cP) 45.9 45.4 45.7 Average energy dissipation (W/kg) 1.65 1.66 2.53

Phase Transition Viscosity bulk volume (cP) 87.6 85.2 52.1 Viscosity near baffles (cP) 7.37 23.6 3.8 Viscosity near impeller (cP) 0.3 0.2 1.12 Average energy dissipation (W/kg) 1.6 1.64 1.34

Final Viscosity bulk volume (cP) 10.1 10 7.87 Viscosity near baffles (cP) 3.5 5.7 2.61 Viscosity near impeller (cP) 0.9 0.7 1.5 Average energy dissipation (W/kg) 1.6 1.64 1.42

Reynolds number for the Time zero material in the RC-1 were in the Laminar region and calculations were done using Visimix ® Laminar. All other calculations done using Visimix ® Turbulent.

o R-203 @ 110 RPM is best matched in R-1002 @ 90 RPM

o Viscosity (across the entire tank), and average energy dissipation values are in good agreement at all three stages of the process that were modeled.

o The RC-1 was able to come close to

matching the viscosity and energy dissipation as compared to the larger equipment

The mixing in 1000 gallon can be adjusted to match that of the 275 gallon.

Data from RC-1 is representative of the large scale equipment.

Example 3: Summary

• As the process changes the mixing can as well

– Physical changes in the process can change how the energy from the impeller is transferred through the material

• Need to compare the mixing at multiple points in the process to be sure the mixing will be sufficient.

Summary

• VisiMix is a valuable tool in the scale up of chemical processes.

• VisiMix helps in the identification of the critical mixing parameters

• VisiMix allows for modeling the mixing of the process at both the large and small scale – Test the process at the small scale – Ensure that the results at the small scale represent the

large scale – May (will) require more than one experiment

• Identify and solve problems before scale…Save $$$!